1. No category

Regime Shifts in the Anthropocene
Juan Carlos Rocha
Stockholm Resilience Centre
Stockholm University
Stockholm, 2015
Para Lucila quien siempre me recordó que resiliencia es la capacidad de salir fortalecido de
las adversidades
To Lucila who always reminded me that resilience is the capacity to persist and learn in the
face of adversity
“Podrán cortar todas las flores, pero no podrán detener la primavera”
Pablo Neruda
2
List of Papers
Paper I:
Biggs, R; Peterson, G & Rocha J.C.
The Regime Shifts Database: A Framework for
Analyzing Regime Shifts in Social-Ecological Systems. [Submitted to Ecology and
Society]
Paper II:
Rocha, J.C.; Peterson, G; & Biggs, R. Regime Shifts in the Anthropocene: Drivers,
Risk and Resilience. [Submitted to Nature]
Paper III:
Rocha J, Yletyinen J, Biggs R, Blenckner T, Peterson G. Marine regime shifts:
drivers and impacts on ecosystems services. Philos Trans R Soc Lond, B, Biol Sci.
The Royal Society; 2015 Jan 5;370(1659):20130273.
Paper IV:
Rocha, J.C. & Wikström, R. Detecting Potential Impacts on Ecosystem Services
Related to Ecological Regime Shifts – A Matter of Wording. [Submitted to Scientific
Reports]
Other publications by the author not included in the thesis:
Book chapter:
Rocha, J.C; Peterson, G & Biggs, R. 2012. Regime Shifts In: Craig et al (eds). Berkshire
Encyclopaedia of Sustainability: Ecosystem Management and Sustainability. Berkshire
Publishing. Barrington, USA.
Contributions to the Regime Shift Database:
Rolands Sadauskis, Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson. Indian
Summer Monsoon. In: Regime Shifts Database, www.regimeshifts.org.
Juan Carlos Rocha, Albert Norström, Reinette (Oonsie) Biggs, Garry Peterson. Coral
Transitions. In: Regime Shifts Database, www.regimeshifts.org.
Matteo Giusti, Christine Hammond, Reinette (Oonsie) Biggs, Garry Peterson, Juan Carlos
Rocha, Brian Walker. Soil Salinization. In: Regime Shifts Database,
www.regimeshifts.org.
Rolands Sadauskis, Garry Peterson, Reinette (Oonsie) Biggs, Juan Carlos Rocha. Arctic
Sea-Ice Loss. In: Regime Shifts Database, www.regimeshifts.org.
Rolands Sadauskis, Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson.
Greenland ice sheet collapse. In: Regime Shifts Database, www.regimeshifts.org.
Rolands Sadauskis, Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson.
Thermohaline circulation. In: Regime Shifts Database, www.regimeshifts.org.
Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson, Rutger Rosenberg. Hypoxia.
In: Regime Shifts Database, www.regimeshifts.org.
Juan Carlos Rocha, Reinette (Oonsie) Biggs, Bob Scholes, Garry Peterson. Bush
Encroachment. In: Regime Shifts Database, www.regimeshifts.org.
3
Christine Hammond, Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson. Bivalves
Collapse. In: Regime Shifts Database, www.regimeshifts.org.
Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson. Kelp Transitions. In: Regime
Shifts Database, www.regimeshifts.org
Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson. Forest to Savannas. In:
Regime Shifts Database, www.regimeshifts.org.
Rolands Sadauskis, Juan Carlos Rocha, Reinette (Oonsie) Biggs, Garry Peterson. Tundra to
Boreal forest. In: Regime Shifts Database, www.regimeshifts.org.
Anneli Sundin, Maja Berggren, Beom-Sik Yoo, Reinette (Oonsie) Biggs, Juan Carlos Rocha.
Sprawling vs Compact City. In: Regime Shifts Database, www.regimeshifts.org.
Johanna Källén, Alba Juárez Bourke, Dayana Hernández Vivas, Kerstin Hultman-Boye, Juan
Carlos Rocha, Reinette (Oonsie) Biggs, Albert Norström, Örjan Bodin. Seagrass
transitions. In: Regime Shifts Database, www.regimeshifts.org.
Juan Carlos Rocha, Reinette (Oonsie) Biggs. Mangroves transitions. In: Regime Shifts
Database, www.regimeshifts.org.
Juan Carlos Rocha, Garry Peterson, Henrik Österblom, Reinette (Oonsie) Biggs. Fisheries
collapse. In: Regime Shifts Database, www.regimeshifts.org.
Susa Niiranen, Garry Peterson, Reinette (Oonsie) Biggs, Juan Carlos Rocha, Henrik
Österblom. Marine food webs: community change and trophic level decline. In:
Regime Shifts Database, www.regimeshifts.org.
Linda Lindström Lindström, Katja Malmborg, Lara D. Mateos, Juan Carlos Rocha, Garry
Peterson. Coniferous to deciduous boreal forest. In: Regime Shifts Database,
www.regimeshifts.org.
Hannah Griffiths, Elinor Holén, Jessica Spijkers, Hannah Griffiths, Elinor Holén, Jessica
Spijkers, Reinette (Oonsie) Biggs, Juan Carlos Rocha. Thermokarst lake to terrestrial
ecosystem. In: Regime Shifts Database, www.regimeshifts.org.
Other online resources:
Rocha, J.C. Regime Shifts. Doctoral essay made available in Wikipedia at:
http://en.wikipedia.org/wiki/Regime_shift
Rocha, J.C. Regime Shifts – Background education material for MOOC on “Planetary
Boundaries and Human Opportunities: The Quest for Safe and just Development on a
Resilience Planet” at: https://www.sdsnedu.org/learn/planetary-boundaries-andhuman-opportunities-fall-2014
Working papers:
Rocha, J.C., Huitric, M & G. Peterson. Assessing Resilience of Arctic Communities to
Social-Ecological Regime Shifts: A Qualitative Comparative Analysis of Adaptive
Capacity
Rocha, J.C. ________. Cascading effects of Regime Shifts: Exploring Hypothesis of
Inconvenient Feedbacks and Domino Effects.
Rocha, J.C., Castillo, D. Misperception of feedbacks: another source of vulnerability in
social-ecological systems.
4
Abstract
Abrupt and persistent reconfiguration of ecosystem’s structure and function has been
observed on a wide variety of ecosystems worldwide. While scientist believe that such
phenomena could become more common and severe in the near future, little is known about
the patterns of regime shifts’ causes and consequences for human well-being. This thesis
aims to assess global patterns of regime shifts in social-ecological systems. A framework for
comparing regime shifts has been developed as well as a public forum for discussing
knowledge about regime shifts, namely the regime shift database. The most common drivers
and expected impacts on ecosystem services have been identified by studying the
qualitative topology of causal networks as well as the statistical properties that explain their
emergent patters. Given that long time series data for ecosystems monitoring is rather
sparse, and experimenting with ecosystems at the scales required to understand their
feedback dynamics is rarely an option; we also proposed an indirect computationally based
method for monitoring changes in ecosystem services. I hope the results here presented
offer useful guidance for managers and policy makers on how to prioritize drivers or impacts
of regime shifts: one take home message is that well-understood variables are not necessary
the ones where most managerial efforts need to be taken. I also hope the scientific
community rigorously criticize our results, but also acknowledge that when doing theoretical
or empirical work, our methods tend to ignore the multi-causal nature of regime shifts. By
bringing back multi-causality to the scientific debate, I hope our results offer new avenues
for hypothesis exploration and theory development on the human endeavour of
understanding Nature.
Resumen
Transiciones críticas o cambios de régimen en ecosistemas se definen como
reconfiguraciones abruptas de su estructura y función. Estos cambios, en ocasiones
inesperados, se han documentado en una gran variedad de ecosistemas en todo el planeta.
Algunos científicos proponen que en el futuro cercano dichos fenómenos pueden volverse
más frecuentes y severos. Sin embargo, sabemos muy poco sobre las causas y
consecuencias potenciales para el bienestar humano. El objetivo de esta tesis es evaluar
patrones globales de cambios de régimen en sistemas socio-ecológicos. Un marco
conceptual para comparar cambios de régimen y un foro público de discusión sobre el
estado del arte en su conocimiento fue desarrollado en la base de datos virtual
www.regimeshifts.org. Las causas más comunes y los impactos en servicios ecosistémicos
más esperados han sido identificados estudiando las propiedades topológicas de redes
causales, así como las propiedades estadísticas que explican sus propiedades emergentes.
Dado que experimentar con ecosistemas a la escala adecuada para capturar sus
mecanismos causales generalmente no es una opción, y dado que la disponibilidad de
datos de largo plazo necesarios para monitorear cambios de régimen son la excepción y no
la regla, proponemos un método indirecto computacional para monitorear cambios en
servicios ecosistémicos. Espero que los resultados sean de utilidad para actores
encargados del diseño de políticas o del manejo de ecosistemas, especialmente espero que
ofrezcan una guía sobre cómo priorizar causas y consecuencias de estos cambios de
régimen: una lección clave es que las variables que mejor entendemos o las que más
monitoreamos no son necesariamente aquellas en las que debemos enfocar las estrategias
de manejo. También espero que la comunidad científica critique con rigor nuestros
resultados, pero a su vez reconozca que tanto el trabajo empírico y teórico como los
métodos que comúnmente se utilizan para estudiar cambios de régimen tienden a ignorar su
naturaleza multi-causal. Al enfatizar la diversidad de sus causas, espero que los resultados
ofrezcan nuevas posibilidades para la exploración de hipótesis y el desarrollo de teorías
para entender mejor la Naturaleza.
5
Sammanfattning
Abrupt och ihållande omkonfigurering av ekosystems struktur och funktion har observerats i
en mängd olika ekosystem världen över. Forskning visar på att dessa fenomen antas bli
vanligare och allvarligare inom vår närmsta framtid. Kunskapen kring dessa s.k. regimskiften
är dock bristfällig, framförallt kring dess konsekvenser för mänskligt välbefinnande. Denna
avhandling syftar till att bedöma globala mönster av regimskiften. Ett ramverk för att jämföra
regimskiften, samt ett offentligt forum, “the regime shifts database”, för att främja diskussion
och sprida kunskap om regimskiften, har utvecklats. De mest förekommande drivkrafter och
effekter på ekosystemtjänster har identifierats genom att studera kvalitativa topologiska och
kausala nätverk, samt de statistiska egenskaperna som förklarar deras framväxande
mönster. Då långvariga tidsserier av ekosystemövervakning är få, och då de experiment som
krävs för att förstå regimskiftens återkopplingsdynamik sällan är möjliga, föreslås också en
indirekt beräkningsmetod för övervakning av förändringar i ekosystemtjänster. Resultaten
från denna avhandling ämnar ger värdefull vägledning för beslutsfattare om
prioriteringsordningen mellan olika typer av drivkrafter och effekter av regimskiften. En viktig
slutsats är att gedigen kunskap om en viss variabel inte nödvändigtvis ger området där
insatser bör tillsättas. Vidare, genom att föra tillbaka multi-kausalitet till den vetenskapliga
debatten, erbjuder avhandlingen nya vägar för hypotesprövning och teoriutveckling inom vår
gemensamma strävan att förstå Naturen.
6
List of Papers
3
Introduction
8
Theory
History of the concept and evidence
Why are regime shifts hard to detect and costly to reverse?
Summary of Results
What are the main drivers of regime shifts?
What can we do to avoid regime shifts?
What are the main impacts on ecosystem services?
8
10
10
12
12
12
13
Summary of methods
13
Conclusions
15
Acknowledgements
16
References
17
7
Introduction
Since the industrial revolution the world’s economy has grown by a factor of 50, energy
consumption by a factor of 40; and since my parents were born (1950s) human population
doubled, all together causing an unprecedented rate of change in world’s ecosystems1,2. The
‘Anthropocene’ has been proposed as a new geological epoch where humans as a collective
have the ability to shape planetary scale processes1. Under current trends of environmental
forcing, scholars suspect that ecosystems will undergo more frequent and severe critical
transitions or regime shifts1,3. Regime shifts are abrupt reorganization of a system’s structure
and function4-6, where a regime corresponds to a characteristic behavior of the system
maintained by mutually reinforcing processes or feedbacks7. The shift occurs when the
strength of such feedbacks change, usually driven by cumulative change in slow variables,
external disturbance or stochastic shocks.
Regime shifts present a challenge for ecological management and governance, because
they are difficult to predict8 and reverse9, while having substantial impacts on the availability
of ecosystems services10. Examples of regime shifts include i) well-established cases such
as freshwater eutrophication11, where lakes turn from clear water to murky water leading to
reduced fishing productivity and toxic algae blooms; ii) controversial cases such as dryland
degradation when dry forest and savanna shift to deserts and bare soils, significantly
reducing ecosystem services such as agricultural production and water cycling12; and iii)
speculative regime shifts such as the melting of the Greenland ice sheet where the frequency
and intensity of warm events will shift the ice sheet from permanent to occasional, reducing
services such as coastline protection and climate regulation13.
Although we have a fairly good theoretical understanding of regime shifts in ecosystems9,14,
little is know about the overall patterns of regime shifts causation and their impacts on
human well being10,15. We do not know how common they are, what could be their long
lasting consequences, where are they most likely to happen, what are their main drivers,
what can we do to avoid them, what are the potential cascading effects or who will be the
most affected groups in society. These questions have inspired most of my work during the
last 6 years. First I will introduce the concepts behind regime shifts theory, then the answers
to my research questions followed by some methodological reflections. The thesis
concludes by revisiting the questions that reminds unanswered, hopefully providing a plan
for future research.
Theory
Regime shifts are large, abrupt, persistence changes in the function and structure of
systems16. They have been documented in a wide range of systems including financial
markets17, climate18, the brain19, social networks20 and ecosystems4,6. Regime shifts are also
known as critical transitions, phase transitions or phase shifts in other disciplines; but I use
the terms interchangeably. Regime shifts in ecosystems are policy relevant because they
can effect the flow of ecosystem services that societies rely upon, they are very difficult to
predict8,21 and often hard to impossible to reverse9,22.
These critical transitions have been documented on a broad range of marine, terrestrial and
polar ecosystems4,14. Two well studied examples include the eutrophication of lakes, when
they turn from clear to murky water affecting fishing productivity and on extreme cases
human health11; or the transition of coral reefs from coral dominated to macro-algae
dominated reefs23,loosing the provision of services related to tourism, coastal protection and
food24-26. Controversial examples include dryland degradation or the shift from forest to
savannas, where we know they can happen but we do not fully understand their
mechanisms and there are disagreements about their drivers12,27,28. Proposed examples of
regime shifts includes the weakening of the Indian Monsoon29,30 or the weakening of the
Thermohaline circulation in the ocean31,32. We know both can happen from paleological
records, but to what extent they are likely under present conditions is less understood.
8
What these phenomena have in common, regardless of scale or system type, is that their
scientific understanding relies on the same mathematical theory of dynamic systems. The
behavior of a system can be described by a set of equations that define all possible values
of their variables, for instance coral cover or fish abundance. These systems tend to
fluctuate around regions of the parameter space that are called equilibrium, basins or
domains of attraction. Systems prone to regime shifts have more than one domain of
attraction; this means that roughly under the same parameter values they can suddenly shift
from one domain to another when critical thresholds are crossed33-35. Note in Figure 1 that
these domains of attraction or regimes are not punctuated equilibriums, they are dynamic,
they fluctuate within the boundaries of the basin of attraction. Thus for example, a forest is
never the same forest twice: the species composition, their abundance and distribution is
always changing as the forest endures fires, hard winters, storms or pests outbreaks. Yet its
identity can always be characterized as a forest. The same can be said of any regime.
120
Forward
Backward
Repeler
100
Vegetation [V]
80
60
40
20
0
−20
0
2
4
6
8
10
Harvesting rate [hmax]
0.00
Density
0.02
0.04
dV
dt
0.06
= rV(1 −
V
V
K
V + hv
) − hmax(
)
0.08
0.10
−20
0
20
40
60
80
100
120
Vegetation [V]
Figure 1. A minimal model of vegetation shifts. The left panel shows a representation of
the bifurcation manifold, this is the parameter space where the system tends to an
equilibrium characterized by forest or dominated by grasslands. Depending of the
parameters of the system it can respond to disturbances in a quasi-linear fashion (back), a
threshold-like response (middle) or present hysteresis (front). On the right panel the
hysteresis case is further elaborated. The system shifts from being dominated by dense
vegetation when the harvesting rate reach values close to 9; however, one has to reduce the
harvesting rate down to 3 for the system to flip back. The probability distribution for the
model behaviour (lower panel) shows a representation of the basins of attraction. Another
way or representing basins of attraction is by ploting the potential function of the system.
The equation that determines the dynamics of this toy model is composed by two terms. The
vegetation dynamics over time is the logistic growth determined by the growth rate r and the
carrying capacity K minus a saturation harvesting term determined by the maximum
harvesting capacity hmax and the harvest h.
9
However, a regime shift does not imply any change of an ecosystem, it has to affect the
feedback structure of the system that maintain its stability and emergent properties7,36; this
is, a change that effects its identity (eg. Forest, coral reef, mangrove). Scientist when they
refer to regime shifts often emphasize that they are abrupt and persistent changes. It worth
noticing that the abruptness and persistence are relative to the dynamics defining the
system’s identity37. Back to the forest example, one generation of trees can last hundreds of
years, then a change over the time span of a decade is abrupt from the forest perspective.
History of the concept and evidence
First notions of systems that can have multiple attractors date back to 1885 when Henri
Poincaré first described bifurcations38. However, only until late 1960s these mathematical
developments were introduced to ecology when leading scientist debated about stability in
ecosystems34,39, which lead to the first theoretical developments, often models, of potential
regime shifts in fisheries, disease outbreaks, and grazing – harvesting systems35,40-42. These
early developments were criticized by the lack of empirical evidence in the 1980s43, but by
early 2000s enough evidence have accumulated on systems such as coral reefs and lakes,
validating the existence of regime shifts and igniting further research on their detection and
management9,14,44.
Nowadays there are still some differences on how the concept is used across disciplinary
fields, especially between community and population ecologist45. For example, while
oceanographers consider a regime shift a phenomenon on the scales of decades46 that has
to be driven by climate, marine biologist focus on the time span of key species life cycles47,48.
Another source of confusion is disciplinary jargon. See for example the contradictory use of
‘phase shifts’ in coral reefs in refs 49 and 50, which is not necessary supported by theory 5.
Evidence of regime shifts has accrued in the last two decades4. It comprises empirical
evidence when experimentation is possible, observational evidence from long-term time
series data, paleological records from sediments and ice cores, as well as mathematical
models that explore the possible mechanisms explaining such observations. Yet, criticism
persist based on the lack of empirical support, ambiguous system boundaries definitions, or
inconsistent disciplinarily jargon36,43,51,52.
Why are regime shifts hard to detect and costly to reverse?
Empirical and theoretical developments have shown that regime shifts are hard to detect
and sometimes hard or impossible to reverse8,9,53. Change in slow variables (e.g. temperature
in coral reefs, nutrients accumulation in lakes) often shrinks the basin of attraction, making
the system more susceptible to shifting when exposed to shock events (e.g. hurricanes,
storms) or the action of external drivers54. When the shift occur the strength of feedbacks
change and some times new feedbacks mechanism come to place, making difficult to
restore the original system state55. This difficulty is often termed ‘hysteresis’; this mean that
the critical tipping point at which the system flip from one regime to another is not the same
at which the system is expected to flip back.
An illustrative iconic example is eutrophication, the shift from clear to murky waters in lakes.
The most common driver of eutrophication is the accumulation of nutrients such as nitrogen
and phosphorous11,56, although it has been reported that it can also by caused by climate
change alone57. Excess of nutrients in the water often come from the use of fertilizers when
growing food. Nutrients are naturally present in the soil and in combination with other
indirect drivers such as deforestation, nutrient inputs both natural and anthropogenic can
accelerate when rain comes and wash the soils. Nutrients accumulate on lake sediments
and are kept by rooted plants. But when a threshold of nutrients on the water is transpassed,
microalgae growth explodes creating algae booms. These booms reduce light penetration to
the bottom of the lake, which reduces the ability of rooted plants to survive. With less rooted
plants the nutrients accumulated on sediments are resuspended on the water column,
reinforcing the original cause of the algae blooms. This is a new reinforcing feedback in
place that makes the regime shift hard to reverse. Figure 2 shows the structure of the system
10
by summarizing the key feedbacks that maintain the clear water regime or the eutrophic
state respectively.
Detecting regime shifts is a challenging task54,58. First one needs to clearly define the
boundaries of the system and select focus variables that serve as indicators of the system
state. Data requirements will depend directly on the spatial and temporal scales at which the
underlying feedback mechanisms operate. Sometimes regime shifts are identifiable from
historical data by looking at jumps on the time series; however statistical techniques are
required to test whether it was a real regime shift. Amongst the most common test used are
the principal component analysis, which compresses several time series of related variables
into few uncorrelated ones; chronological clustering, and sequential t-tests59,60. Recent
efforts on improving prediction rather than detection from past events have been focusing
on the development of early warning signals54. They are based on the statistical signature of
a system approaching a regime shift called critical slowing down which can be observed on
the increased autocorrelation (temporal or spatial) and skewness while decrease in
variance61-64.
Fishing
predatory fish+
+
forage fish
B
-
P recycling mechanism is a
process containing two
feedbacks: one of the
macrophytes increasing P in
water, other throuhg DO
increasing P in water
B
R
-
Turbidity
-
+
Urbanization
Macrophytes
Sewage
R
DO
-
-
Algae
+
+
B
Urban storm
water runoff
-P in Water
+
-
+
+
+
Nutrients input
+
+
+
R
Temperature
- Flushing
+
-
+
Fertilizers use
Droughts
+
+
Floods
+
Climate Change
+
irrigation
+
+
+
Impoundments
+
Rainfall variability
Erosion +
Agriculture
+
Deforestation
Greenhouse
gases
Figure 2. An illustrative example of freshwater eutrophication. The causal loop diagram
summarizes the potential feedback loops (coloured) underlying the dynamics. In black are
the direct and indirect drivers of eutrophication. P stands for phosphorous and DO is
dissolved oxygen.
Research on regime shifts has adopted either an empirical, modelling11 or early warning
indicator approach54,58, which requires deep knowledge of the causal structure of the system
or high quality spatial-temporal data respectively. These requirements have largely confined
regime shift research to the analysis of individual regime shifts rather than their comparison,
with a handful of exceptions for climate18, agriculture65, and hydrology66 related regime shifts.
But what do we do when we do not have the luxury of long term monitoring programs or
incomplete understanding of underlying mechanisms? How do we inform managers about
the risk of critical transitions with impacts to society? Can we overlay knowledge from a
system to another to better understand patterns of causation and potential impacts on
human well-being?
11
Summary of Results
This thesis attempts to fill these gaps by assessing global patterns of regime shifts. We have
developed an analytical framework to study and compare regime shifts across different
systems and scales (Paper I). The Regime Shifts Database www.regimeshifts.org is an effort
to systematically collect, synthesize and communicate current knowledge about regime
shifts. Our initial focus is regime shifts that matter to people, that have potential impacts on
ecosystem services and human well-being. Our database inherits the focus on non-linear
dynamics from its predecessor, the thresholds database developed by the Resilience
Alliance and Santa Fe Institute67. Our database, however, advances the conceptual
framework to enhance comparability across cases by stressing the role of feedback
mechanisms, distinguishing drivers from processes and highlighting impacts on ecosystem
services as well as management options. Thanks to the collaboration of over 60 contributors
(researchers and grad students), the database offers today synthesis of >800 scientific
papers, summarizing more than 200 cases and about 30 generic types of regime shifts
across the globe.
What are the main drivers of regime shifts?
Drivers are defined as any natural or human-induced factor that directly or indirectly causes
a change in an ecosystem68. Of course, that depends on the imaginary boundaries that we
pose on reality when delimiting the ‘system’ of analysis. While those boundaries change
from observer to observer and from one scientific discipline to another; understanding the
relative importance of drivers requires embracing their diversity.
Inspired by recent applications of complex systems science to the network of human
diseases69,70, we have applied similar methods to disentangle the main drivers of regime
shifts and their impacts on ecosystem services in settings where very limited information
regarding field data is available (time series or spatial data), or when evidence about the
causal relationships underlying their mechanisms is contested (Papers II & III).
Although most studies focus on few drivers, we show that regime shifts are multi-causal
phenomena and the co-occurrence patterns of drivers can inform managerial strategies. For
marine ecosystems (Paper II) we found that food production related drivers, coastal
development and climate change are the most important drivers. Except by coastal
development, this pattern holds when we scale up the analysis to the globe (Paper III). In
fact, these drivers are thought to reinforce each other71-73 increasing the risk of cross-scale
interactions and cascading effects74,75.
What can we do to avoid regime shifts?
Not only the diverse scales of key drivers, but also the high co-occurrence of drivers across
scales means that managing regime shifts requires coordinating actions, more often than
expected at the international scale (Paper III). While marine regime shifts share significantly
more drivers suggesting high similarity on their underlying mechanisms, terrestrial regime
shifts share fewer drivers making managerial options more context dependent (Papers II &
III). Our results show that most management options required to tackle drivers of regime
shifts require international coordinated actions, hopefully they offer guidance on the
feedbacks that institutions needs to match to avoid the problem of fit76.
However, by managing local to regional drivers, managers can build resilience to regime
shifts and delay the impact of global ones. Our results suggest that in situations where
regime shifts dynamics are poorly understood, managerial options that work in well
understood regime shifts can be applied on uncertain data scarce cases if they share similar
ecosystem processes, occur in similar ecosystems and whose dynamics occur at similar
spatio-temporal scales (Paper III). A deeper analysis on marine regime shifts further forewarn
that management for well understood or well monitored variables might not be enough to
preclude regime shifts from happening (Paper II). Drivers diversity on one hand, and the
different causal pathways they can take to affect feedback mechanism on the other, suggest
that management should embrace the multi-causal nature of regime shifts and address more
12
often indirect drivers even though it is harder to get statistical signature of they causal role
on the regime shift, as more intermediate steps between cause and effects take place.
What are the main impacts on ecosystem services?
Our network analysis suggests that biodiversity loss, food production (fisheries, primary
production, nutrient cycling), disease control and aesthetic values are the ecosystem
services most commonly impacted by regime shifts globally (Papers II & III). However,
predicting which ecosystem services are likely to be affected by ecological surprises is still
one of the greatest challenges of current ecological research10. Given that experimenting
with large-scale ecosystems is rarely an option, in joint efforts with computer scientists we
explore an indirect approach by using machine learning algorithms for topic modeling in the
scientific literature (Paper IV). We used the Millennium Ecosystem Assessment as training
dataset and a corpus of >800 papers about regime shifts. We found that identifying
ecosystem services is possible; the algorithm identifies fairly well provisioning and regulating
services but finds more difficulties to correctly identify supporting and cultural ones. We
hope our results will open an avenue for using social media as environmental sensors of
ecosystem change.
Summary of methods
The previous section outlined the main research questions of the thesis as well as the key
findings that each paper addressed. For the academic inclined reader, this section will
present how questions and findings meet through the application of scientific methods. Fist I
will present a common denominator through my work, something I call the size of the
problem or more technically, the parameter space that is to be explored. To better explain it I
will use as metaphor the case of cancer to illustrate lessons from biomedical research that
can be applied to the problem of regime shifts. Then I justify my methods selection and
reflect on their limitations.
Not all regime shifts occur in ecosystems, a regime shift more familiar to the average reader
is cancer77. Unfortunately everyone knows a relative, colleague or acquaintance that have
suffered of cancer. Cancer is in essence a genetic disease 78. While the human genome is
thought of being composed by ~30.000 genes, cancer is know to be caused by ~547 genes
79
, with some of them playing a more important role than others77. For some of those gene
mutations we know exactly what causal pathway it cascades and what are the potential
effects on cellular development; however, for others we only have correlational reports of
patients having mutations or atypical concentrations of substances that the gene codify for,
but we do not know the mechanisms in place that can related to cancer 78. In an ideal world
where scientist could experiment with humans to explore the problem space of cancer, they
will have 2547 = 4.6e10164 experimental combinations –assuming order of mutation does not
matter; which plus replicates (or without) rapidly demonstrates that we are not enough
humans to run the experiment. Moreover, cancer is an evolutive disease, it can adapt to
treatments and such adaptation varies from patient to patient. How do we crack that
problem space? How do we advance knowledge without the luxury of rigorous experiments
and an uneven understanding of the role of the causal mechanism?
In social-ecological regime shifts the problem is the same, although the boundaries of the
systems are not as clear as with cancer; and the knowledge database is definitely less
developed. We have identified ~54 drivers of regime shifts, a list that without doubt is far
from complete. As with cancer, for some drivers we know fairly well the mechanism in place
although the strength change from place to place (e.g. deforestation), for some other drivers
knowledge is contested and evidence inconclusive (e.g. poverty or trade). On an ideal world
of free scientific experimentation without ethical concerns, we do need more than one planet
to rigorously test for significance. No need to use the ~54 drivers for calculation this time
(254= 2916 combinations); the caveat here is the unit of analysis, the system boundary. Take
as example the transition from forest to savanna, it has 13 reported drivers (213= 8192
combinations), but we only have few rainforest patches to experiment with moisture
recycling feedbacks and the extent of fires (Amazon, Congo basin, Malaysia, Indonesia and
13
Papua New Guinea). When it comes to regime shifts, long time series data and the
possibility of experimentation is the exception rather than the rule.
The first step on the quest of assessing global patterns of regime shifts was creating a
platform for comparison (Paper I). As medical doctors have maps of metabolic pathways, we
constructed maps about the causation narratives described by scientific papers on the form
of causal loop diagrams. Causal loop diagrams (CLDs) are a technique to map out the
dynamic structure of a system80,81, they are known to have limitations and obscuring
accumulation processes82, yet they are excellent tools for communication and education81.
CLDs were key for distinguishing drivers from feedbacks, hypothesize the boundaries of the
system in question, and collect knowledge that otherwise is fragmented across academic
disciplines. The next step was networks.
Why networks? Network science has proven to be a versatile tool for understanding the
structure and functioning of complex systems 83, particularly when one needs to understand
emergent systemic patterns from localized interactions. The list of applications of network
theory to different fields in science is endless, but to keep it short, one of my sources of
inspiration was the application of networks to the mapping of human diseases because of
the striking similarities of the problem size above described. Networks of genes interactions,
proteins, diseases and symptoms are revolutionizing the way we understand medicine
today69,70,84-87. Can we say the same about ecosystem management? By studying the
statistical properties of the observed networks as oppose to random counterparts we could
identify which drivers tend to co-occur more often than expected and which ecosystem
services are more impacted by regime shifts. Exponential random graph models 88-90 were
deployed on Paper II to test for how local interactions explained or not the observed global
patterns.
Attracted to methods to explore a large parameter space with practical applications, I
become interested on latent Dirichlet allocation (LDA). Imagine in how many different ways
one can combine words to convey the same meanings, yet if you are talking about a specific
topic, say soccer, the frequency distributions of the words selected is probably not random.
Paper IV is in itself the exploration of this machine learning technique (LDA) as an indirect
method to monitor changes in ecosystem services that in turn could indicate potential
regime shifts. Real world applications for decision making in ecosystem management and
sustainability science are still an unexplored field.
Table 1. Summary of methods
Question / Method
How to compare
regime shifts?
What are the main
drivers of regime
shifts?
What can we do to
avoid regime shifts?
What are the main
impacts on ecosystem
services?
Causal loop diagrams
Paper I: framework
development
Networks
Topic mining
Papers II & III:
Networks simulations,
exponential random
graph models,
clustering and
multidimensional
scaling
Paper IV: Latent
Dirichlet allocation.
The selection of methods for this thesis matched the type of questions asked and the data
available at the global scale to answer them. However, all methods have limitations that
constraint the extent to which the conclusions stand by themselves. Causal loop diagrams
are excellent tools when mapping different plausible explanations of why and how regime
shifts happen; they facilitate communication and the use of systems thinking, they force you
to be clear on what is consider exogenous, endogenous and the types of causal
relationships. However, they are still a subjective conceptual interpretation of what otherwise
is a scientific narrative, they are very hard to defend as source of data since replication is
context dependent. Latent Dirichlet allocation is a method that has been recently criticized
because of reproducibility and accuracy issues91, problems that we did encounter when
14
searching for ecosystem services. In addition, LDA is a method designed as an explorative
tool that we forced into prediction. Nevertheless, that was the purpose of Paper IV: exploring
when does it works, when it does not and why. Potential methods that would outperform
LDA on topic detection and classification employ community detection algorithms in
networks91,92, an avenue still to be explored in our dataset.
Networks are a useful framework to study complex systems, yet they have been criticized by
issues of scalability and lack of dynamics93. Thus far, the first critique does not affect any of
our datasets since we do not handle millions of nodes. The second, however, does limit the
conclusions of my studies since they are all based on structural patterns of a network, not its
dynamics. For example, the weighting of links between nodes have been based on cooccurrence. As the writing of this thesis, there is not enough data that allows one to use as
weights the relative contribution of a driver to a regime shift at the global scale; moreover, it
is expected that spatial heterogeneity94 or the ordering of stress events would change the
relative weights of drivers. Dynamic modeling across many case studies and empirical
studies would be necessary to advance knowledge in that direction. However, our results
are a qualitative characterization of potential drivers interactions that offer a fertile ground for
generating hypothesis, guide the development of dynamic models and data collection. I
believe our conclusions are robust to different choices of categorization (e.g. which drivers
to include or not). In other words, it is likely that we have missed regime shifts cases, drivers
or impacts on ecosystem services; but it is not likely that including them will substantially
change the patterns here reported.
Conclusions
The Anthropocene is pushing humanity outside its comfort zone, an arena where critical
transitions in ecosystems are likely to become more frequent and severe. How do we
prepare, how do we navigate uncertainty with incomplete knowledge about their dynamics,
how do we provide scientific advise when experimenting is rarely an option and waiting for
observation might not be socially affordable? This thesis tried to meet the challenge by
assessing global patterns of regime shifts: what they have in common, and identifying what
can we learn from well-understood cases that can be applicable to more speculative ones.
We have developed a conceptual framework to facilitate comparison and knowledge
synthesis of regime shifts. The framework is both a source of data for future analysis and a
public discussion forum where other scientist, practitioners and managers can contribute
with new knowledge. I hope this open collective effort break the disciplinary silos on which
regime shift research most commonly occur. Methodologically, this is probably the first time
that tools from network analysis are used to disentangle causal patterns on large-scale
phenomena such as regime shifts. Networks have been used in ecology to understand
species interactions (e.g. mutualism, food webs) or landscape dynamics (e.g. fragmentation);
but to my knowledge, this is the first time that networks are used to study the structure of
processes. Topic modeling is also a recently developed technique that is here applied as
novel method to explore ecological problems. Both methods have caveats and limitations,
but have been essential to push knowledge forwards in ways I did not imagine when started
my PhD.
Perhaps the most important lesson learnt is that regime shifts are multi-causal phenomena.
Contrary to the text book examples where the authors usually presents you to ‘the’ driver
and ‘the’ slow variable; reality seems pervaded by multiple ways in which regime shifts
develop, all occurring at the same time. The multidimensionality of these non-linear
phenomena does not seems to be appropriated by scholars in many disciplines that insist on
getting rid of co-linear variables on the search for clean p-values. That is fine for the
statistics in the paper, but in my opinion it misleads the practical application of the science.
As George Sugihara would put it, do no use linear methods to study non-linear phenomena,
ecosystems are complex systems where variables are often mutually coupled creating
feedbacks, in other words, they are not completely independent95.
15
At first drivers diversity might sound discouraging, but we demonstrate that they do not cooccur at random highlighting potential managerial strategies. Marine regime shifts share
more drivers and processes, by targeting a handful of drivers many of these regime shifts
can be prevented. It did not come as a surprise that climate and food production related
drivers are core set of regime shift’s causes worldwide, both topics are already hot debates
on both the scientific and political realms. Yet, the results show that managing local to
regional drivers provide opportunities to delay the effects of global drivers. Surprisingly, the
fact that many co-occurring drivers are indirect support the idea of cascading effects: first it
suggest that risk of regime shifts can increase in sync; second, it supports the possibility of
inconvenient feedbacks through teleconnections. This ideas are further reinforced when
looking from the impacts perspective: subcontinental regime shifts consistently impact
climate regulation, terrestrial regime shifts commonly impact water cycling, food production,
and fresh water, while marine regime shifts typically affect fisheries, water purification,
disease control and surprisingly aesthetic values. It is not hard to imagine how the effects of
a regime shift can become the causes of another elsewhere.
The quest for better understanding cascading effects is at the frontiers of regime shifts
research37,54,75,94,96,97. Imagine for example how the shift from forest to savanna in substantial
parts of the Amazon rainforest will weaken the moisture recycling feedback, reducing rainfall
in the Andes and the ocean. How many regime shifts would be interconnected through
precipitation? Or imagine the weakening of the albedo feedback related to regime shifts in
the Arctic, Greenland and Antarctica ice sheets; which regime shifts would increase the
likelihood of occurring down the road by altering climate? I started exploring this question
during my master thesis: ‘The domino effect: A network analysis of regime shifts drivers and
causal pathways’98. However, methodological challenges have left this work unfinished, for
now. We are not too far of developing a rigorous framework to explore the variety of
hypothetical connections between regime shifts that can inform future theory and empirical
developments on the quest of cascading effects.
Acknowledgements
This work would have not been possible without the guidance and support of my
supervisors Garry Peterson and Oonsie Biggs. Both shaped my thinking, sharped my
questions, and opportunely challenged my assumptions. I enormously appreciate working
with them on a project where knowledge creation was not only for the sake of its scientific
value; but also, to be made available for managers, practitioners and people who need it.
Thanks for patiently dealing with my stubbornness and broken English, thanks for your
support on moments of crises, thanks for giving me the freedom to try out different angles to
tackle my questions and for believing in me when I failed to do so.
I also have to say ‘gracias’ to my live companions and friends who reminded me how much
fun there is outside the office when I got lost on my own thoughts. Gracias to my family for
encouraging me to follow my dreams and adventures. Thanks to Giesela for her smile and
love. Thanks to la abuela Patricia for helping me find perspective. Thanks Axel, Steve, Ingo,
Susa, Jennifer for climbing afternoons and kayaking trips. In addition, thanks Matteo,
Megan, Julia, Kari, Christian, Johan, Jonas, Daniel, Caro, Emilie, Rolands, Johanna,
Vanessa, Gavin, Simon, Diego, Quentin for sushi nights, game nights, parties, hiking and
making my life in Sweden so enjoyable. Thanks to the students of the resilience research
school for inspiring seminars and other academic distractions.
I also would like to acknowledge people and organizations that enriched with their
experience my own academic journey. Gracias Jordi Bascompte for inviting me to visit your
lab and learn from your group how network science can help us unravel the mysteries of
biodiversity, thanks for your feedback, inspiration and a bit of magic. Thanks to Fariba Karimi
and René Pfitzner for organizing and inviting me to the Network Factory summer school, it
was encouraging to see how the language of networks has so many different applications.
Thanks to Uno Svedin for inviting me to participate on the Nordic Consortium for Systems
Analysis (NORCOSA), his course on systems analysis was where the ideas for paper IV were
16
initially conceived, and thanks to Robin Wikström for his support all the way to the end.
Thanks to Nick Magliocca at SESYNC for inviting me to the workshop on social media and
environmental synthesis, the discussions helped me mature my ideas on the possibilities
and limits of topic mining for monitoring ecosystem services. Egbert van Nes and Marten
Scheffer not only organized an amazing course on critical transitions, they have substantially
developed and tested the theory on which the ideas of regime shifts rest, their work has
been a continuous source of inspiration. Thanks especially to Marten for encouraging me to
take daily walks on the forest, or searching the space for associative thinking. Thanks to
Therese Lindahl for inviting me to the workshop on behavioral economics and nature
network, for inviting me to write grant proposals to think of regime shifts from a more human
based perspective. Thanks to Jonas Hentati-Sundberg for thought provoking questions
regarding the existence of regime shifts when looked from the critical empirical eyes of the
non-linear dynamic methods. Marcus Carson, Garry Peterson, Annika Nilsson, Sarah Cornell
and Miriam Huitric thanks for working together towards the resilience assessment of the
Arctic, you have brought a different dimension to my work and have helped me better
understanding how the science-policy interphase happens in real life. Thanks to the
Complex Systems Society and the Swedish Secretariat for Environmental Earth and Systems
Sciences for supporting my participation in several international conferences. Most of my
research was supported by the Formas grant 2009-6966-139149-41.
Last but not less, I could not be more grateful to the contributors of the regime shifts
database, particularly the five generations of master students with whom I had the privilege
of discussing literature on regime shifts. Their questions helped me clarify my own
definitions, their enthusiasm encouraged me to further explore different cases, and their
point of view enriched many of the contributions available today in the database. And to the
Stockholm Resilience Centre for being such a stimulating working place, full of creative
thinkers, visiting researchers and more activities that one can actually fit on a calendar;
thanks for being such an amazing group of people.
References
1.
Steffen, W., Crutzen, P. J. & McNeill, J. R. The Anthropocene: Are humans now overwhelming
the great forces of nature. Ambio 36, 614–621 (2007).
Ellis, E. C., Goldewijk, K. K., Siebert, S., Lightman, D. & Ramankutty, N. Anthropogenic
transformation of the biomes, 1700 to 2000. Global Ecology and Biogeography 19, 589–606
(2010).
Rockström, J. et al. A safe operating space for humanity. Nature 461, 472–475 (2009).
Scheffer, M. Critical Transitions in Nature and Society. (Princeton University Press, 2009).
Solé, R. V. Phase Transitions. (Princeton University Press, 2011).
Solé, R. & Bascompte, J. Self-organization in complex ecosystems. (2006).
Norberg, J. & Cumming, G. S. Complexity Theory for a Sustainable Future. (Columbia
University Press, 2008).
Hastings, A. & Wysham, D. B. Regime shifts in ecological systems can occur with no warning.
Ecol Lett 13, 464–472 (2010).
Scheffer, M. & Carpenter, S. Catastrophic regime shifts in ecosystems: linking theory to
observation. Trends Ecol Evol 18, 648–656 (2003).
Carpenter, S. R. et al. Science for managing ecosystem services: Beyond the Millennium
Ecosystem Assessment. P Natl Acad Sci Usa 106, 1305–1312 (2009).
Carpenter, S. R. Regime shifts in lake ecosystems. (Ecology Institute, 2003).
Geist, H. & Lambin, E. Dynamic causal patterns of desertification. BioScience 54, 817–829
(2004).
Greene, C. H., Pershing, A. J., Cronin, T. M. & Ceci, N. Arctic Climate Change and its Impacts
on the Ecology of the North Atlantic. Ecology 89, S24–S38 (2008).
Scheffer, M., Carpenter, S., Foley, J., Folke, C. & Walker, B. Catastrophic shifts in ecosystems.
Nature 413, 591–596 (2001).
Millennium Ecosystem Assessment. Ecosystems and human well-being: synthesis. (Island
Press, 2005).
Biggs, R., Carpenter, S. R. & Brock, W. A. Turning back from the brink: Detecting an impending
regime shift in time to avert it. P Natl Acad Sci Usa 106, 826–831 (2009).
Choi, Y. & Douady, R. Financial Crisis and Contagion: A Dynamical Systems Approach. SSRN
Journal (2011). doi:10.2139/ssrn.1733706
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
17
Lenton, T. et al. Tipping elements in the Earth's climate system. P Natl Acad Sci Usa 105, 1786
(2008).
Scheffer, M., van den Berg, A. & Ferrari, M. D. Migraine Strikes as Neuronal Excitability
Reaches a Tipping Point. PLoS ONE 8, e72514 (2013).
Christakis, N. & Fowler, J. Social Network Sensors for Early Detection of Contagious
Outbreaks. PLoS ONE 5, e12948 (2010).
Boettiger, C. & Hastings, A. Early warning signals and the prosecutor's fallacy. P R Soc B 279,
4734–4739 (2012).
Suding, K. N. & Hobbs, R. J. Threshold models in restoration and conservation: a developing
framework. Trends Ecol Evol 24, 271–279 (2009).
Nyström, M., Folke, C. & MOBERG, F. Coral reef disturbance and resilience in a humandominated environment. Trends Ecol Evol 15, 413–417 (2000).
Graham, N. et al. Managing resilience to reverse phase shifts in coral reefs. FRONTIERS IN
ECOLOGY (2013). doi:10.1890/120305
Bellwood, D., Hughes, T., Folke, C. & Nyström, M. Confronting the coral reef crisis. Nature 429,
827–833 (2004).
Moberg, F. & Folke, C. Ecological goods and services of coral reef ecosystems. Ecol Econ 29,
215–233 (1999).
Oyama, M. & Nobre, C. Climatic consequences of a large-scale desertification in northeast
Brazil: A GCM simulation study. J Climate 17, 3203–3213 (2004).
Da Silva, R., Werth, D. & Avissar, R. Regional impacts of future land-cover changes on the
amazon basin wet-season climate. J Climate 21, 1153–1170 (2008).
Dixit, Y., Hodell, D. A. & Petrie, C. A. Abrupt weakening of the summer monsoon in northwest
India 4100 yr ago. Geology (2014). doi:10.1130/G35236.1
Malik, N., Bookhagen, B., Marwan, N. & Kurths, J. Analysis of spatial and temporal extreme
monsoonal rainfall over South Asia using complex networks. Clim Dynam 39, 971–987 (2011).
Broecker, W. S. Thermohaline Circulation, the Achilles Heel of Our Climate System: Will ManMade CO2 Upset the Current Balance? Science 278, 1582–1588 (1997).
STOMMEL, H. Thermohaline Convection with Two Stable Regimes of Flow. Tellus 13, 224–230
(1961).
Lewontin, R. C. Meaning of Stability. Brookhaven Sym Biol 13–& (1969).
Holling, C. S. Resilience and stability of ecological systems. Annual review of ecology and
systematics 4, 1–23 (1973).
May, R. Thresholds and breakpoints in ecosystems with a multiplicity of stable states. Nature
269, 471–477 (1977).
Schröder, A., Persson, L. & De Roos, A. M. Direct experimental evidence for alternative stable
states: a review. Oikos 110, 3–19 (2005).
Hughes, T. P., Carpenter, S., Rockström, J., Scheffer, M. & Walker, B. Multiscale regime shifts
and planetary boundaries. Trends Ecol Evol 28, 389–395 (2013).
Poincare, H. Sur l'équilibre d“une masse fluide animée d”un mouvement de rotation. Acta
Math. 7, 259–380 (1885).
Lewontin, R. C. Meaning of Stability. Brookhaven Sym Biol 13–& (1969).
Ludwig, D., Jones, C. G. & Holling, C. S. Qualitative Analysis of Insect Outbreak Systems: The
Spruce budworm and Forest. J. Anim. Ecol. 47, 315–332 (1978).
Noy-Meir, I. Stability of grazing systems: an application of predator-prey graphs. The Journal of
Ecology (1975).
Jones, D. D. & Walters, C. J. Catastrophe theory and fisheries regulation. Journal of the
Fisheries Board of … (1976).
Connell, J. H. & Sousa, W. P. On the evidence needed to judge ecological stability or
persistence. Am Nat (1983).
Folke, C. et al. Regime shifts, resilience, and biodiversity in ecosystem management. Annu Rev
Ecol Evol S 35, 557–581 (2004).
Beisner, B., Haydon, D. & Cuddington, K. Alternative stable states in ecology. FRONTIERS IN
ECOLOGY 1, 376–382 (2003).
Collie, J., Richardson, K. & Steele, J. Regime shifts: can ecological theory illuminate the
mechanisms? Progress in Oceanography 60, 281–302 (2004).
Norström, A., Nyström, M., Lokrantz, J. & Folke, C. Alternative states on coral reefs: beyond
coral–macroalgal phase shifts. Mar Ecol-Prog Ser 376, 295–306 (2009).
Conversi, A. et al. A holistic view of marine regime shifts. Philos. Trans. R. Soc. Lond., B, Biol.
Sci. 370, 20130279–20130279 (2015).
Dudgeon, S. R., Aronson, R. B., Bruno, J. F. & Precht, W. F. Phase shifts and stable states on
coral reefs. Mar Ecol-Prog Ser 413, 201–216 (2010).
DONE, T. PHASE-SHIFTS IN CORAL-REEF COMMUNITIES AND THEIR ECOLOGICAL
SIGNIFICANCE. Hydrobiologia 247, 121–132 (1992).
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
18
51.
Bruno, J. F., Sweatman, H., Precht, W. F., Selig, E. R. & Schutte, V. G. W. Assessing evidence
of phase shifts from coral to macroalgal dominance on coral reefs. Ecology 90, 1478–1484
(2009).
Capon, S. J. et al. Regime shifts, thresholds and multiple stable states in freshwater
ecosystems; a critical appraisal of the evidence. Science of The Total Environment (2015).
doi:10.1016/j.scitotenv.2015.02.045
Boettiger, C. & Hastings, A. Tipping points: From patterns to predictions. Nature 493, 157–158
(2013).
Scheffer, M. et al. Anticipating Critical Transitions. Science 338, 344–348 (2012).
Nyström, M. et al. Confronting Feedbacks of Degraded Marine Ecosystems. Ecosystems
(2012). doi:10.1007/s10021-012-9530-6
Carpenter, S., Ludwig, D. & Brock, W. Management of eutrophication for lakes subject to
potentially irreversible change. Ecol Appl 9, 751–771 (1999).
Winder, M. Limnology: Lake warming mimics fertilization. Nature Climate Change 2, 771–772
(2012).
Scheffer, M. et al. Early-warning signals for critical transitions. Nature 461, 53–59 (2009).
Andersen, T., Carstensen, J., Hernandez-Garcia, E. & Duarte, C. M. Ecological thresholds and
regime shifts: approaches to identification. Trends Ecol Evol 24, 49–57 (2009).
Sonderegger, D. L., Wang, H., Clements, W. H. & Noon, B. R. Using SiZer to detect thresholds
in ecological data. http://dx.doi.org/10.1890/070179 7, 190–195 ( Ecological Society of
America, 2008).
Dakos, V. et al. Methods for Detecting Early Warnings of Critical Transitions in Time Series
Illustrated Using Simulated Ecological Data. PLoS ONE 7, e41010 (2012).
Dakos, V. & Bascompte, J. Critical slowing down as early warning for the onset of collapse in
mutualistic communities. P Natl Acad Sci Usa 201406326 (2014).
doi:10.1073/pnas.1406326111
Dakos, V., Carpenter, S. R., van Nes, E. H. & Scheffer, M. Resilience indicators: prospects and
limitations for early warnings of regime shifts. Philos. Trans. R. Soc. Lond., B, Biol. Sci. 370,
20130263–20130263 (2015).
Dakos, V., van Nes, E. H., Donangelo, R., Fort, H. & Scheffer, M. Spatial correlation as leading
indicator of catastrophic shifts. Theor Ecol 3, 163–174 (2010).
Gordon, L. J., Peterson, G. D. & Bennett, E. M. Agricultural modifications of hydrological flows
create ecological surprises. Trends Ecol Evol 23, 211–219 (2008).
Karlsson, J. M., Bring, A., Peterson, G. D., Gordon, L. J. & Destouni, G. Opportunities and
limitations to detect climate-related regime shifts in inland Arctic ecosystems through ecohydrological monitoring. EPL (Europhysics Letters) 6, 014015 (2011).
Walker, B. & Meyers, J. Thresholds in ecological and socialecological systems: a developing
database. Ecol Soc 9, 3 (2004).
Nelson, G. C. et al. Anthropogenic drivers of ecosystem change: an overview. Ecol Soc 11,
(2006).
Barrenas, F., Chavali, S., Holme, P., Mobini, R. & Benson, M. Network Properties of Complex
Human Disease Genes Identified through Genome-Wide Association Studies. PLoS ONE 4,
e8090– (2009).
Goh, K.-I. et al. The human disease network. P Natl Acad Sci Usa 104, 8685–8690 (2007).
Moore, F. C. & Lobell, D. B. The fingerprint of climate trends on European crop yields. P Natl
Acad Sci Usa 112, 2670–2675 (2015).
Lawrence, D. & Vandecar, K. Effects of tropical deforestation on climate and agriculture. Nature
Climate Change 5, 27–36 (2015).
Ward, D. S., Mahowald, N. M. & Kloster, S. Potential climate forcing of land use and land cover
change. Atmos. Chem. Phys. 14, 12701–12724 (2014).
Debra P. C. Peters et al. Cross-scale interactions, nonlinearities, and forecasting catastrophic
events. P Natl Acad Sci Usa 101, 15130–15135 (2004).
Peters, D. P. C., Sala, O. E., Allen, C. D., Covich, A. & Brunson, M. Cascading events in linked
ecological and socioeconomic systems. FRONTIERS IN ECOLOGY 5, 221–224 (2007).
Young, O. R., King, L. A. & Schroeder, H. Institutions and environmental change. (The MIT
Press, 2008).
Li, C. & Wang, J. Quantifying the underlying landscape and paths of cancer. J. R. Soc.
Interface 11, 20140774–20140774 (2014).
Vogelstein, B. & Kinzler, K. W. Cancer genes and the pathways they control. Nature Medicine
10, 789–799 (2004).
Futreal, P. A. et al. A census of human cancer genes. Nat Rev Cancer 4, 177–183 (2004).
Sterman, J. Business Dynamics: Systems Thinking and Modeling for a Complex World.
(McGraw-Hill/Irwin, 2000).
Lane, D. The emergence and use of diagramming in system dynamics: a critical account.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
19
Systems Research and Behavioral Science 25, 3–23 (2008).
Richardson, G. Problems with causal-loop diagrams. System Dynamics Review 2, 158–170
(1986).
Barabasi, A.-L. Network science: Luck or reason. Nature – (2012). doi:10.1038/nature11486
Zhou, X., Menche, J., Barabasi, A.-L. & Sharma, A. Human symptoms-disease network. Nature
Communications 5, 4212 (2014).
Barzel, B. & Barabasi, A.-L. Network link prediction by global silencing of indirect correlations.
Nature biotechnology 31, 720–725 (2013).
Yildirim, M., Goh, K., Cusick, M., Barabasi, A. & Vidal, M. Drug-target network. Nature
biotechnology 25, 1119 (2007).
Vidal, M., Cusick, M. & Barabási, A. Interactome Networks and Human Disease. Cell 144, 986–
998 (2011).
Krivitsky, P. N. Exponential-family random graph models for valued networks. Electronic
Journal of Statistics 6, 1100–1128 (2012).
Wang, P., Pattison, P. & Robins, G. Exponential random graph model specifications for
bipartite networks—A dependence hierarchy. Social networks 35, 211–222 (2013).
Robins, G., Snijders, T., Wang, P., Handcock, M. & Pattison, P. Recent developments in
exponential random graph ( p*) models for social networks. Social networks 29, 192–215
(2007).
Lancichinetti, A. et al. High-Reproducibility and High-Accuracy Method for Automated Topic
Classification. Phys. Rev. X 5, 011007 (2015).
Gopalan, P. K. & Blei, D. M. Efficient discovery of overlapping communities in massive
networks. P Natl Acad Sci Usa 110, 14534–14539 (2013).
Holland, J. H. Signals and Boundaries. (MIT Press, 2012).
Brook, B. W., Ellis, E. C., Perring, M. P., Mackay, A. W. & Blomqvist, L. Does the terrestrial
biosphere have planetary tipping points? Trends Ecol Evol 28, 396–401 (2013).
Sugihara, G. et al. Detecting Causality in Complex Ecosystems. Science (2012).
doi:10.1126/science.1227079
Lenton, T. M. & Williams, H. T. P. On the origin of planetary-scale tipping points. Trends Ecol
Evol 28, 380–382 (2013).
Kinzig, A. et al. Resilience and regime shifts: assessing cascading effects. Ecol Soc 11, 20
(2006).
Rocha, J. C. The domino effect: a network analysis of regime shifts drivers and causal
pathways. 1–119 (2010).
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.
20